Are neurostimulators implanted within the brain a better approach in epilepsy than less-invasive vagus nerve stimulation (VNS)? That depends, a researcher said.
Real-world data from two large electronic health records databases showed that short-term complications were vastly greater with deep brain stimulation (DBS) or responsive nerve stimulation (RNS), both of which involve threading wires into the brain, according to Samhitha Rai, BS, an MD candidate at Case Western Reserve University and the Cleveland Clinic in Ohio.
But long-term mortality -- which her group considered a proxy for seizure control -- became substantially worse for patients treated with VNS compared with the intracranial implants, starting at about 5 years after surgery, Rai reported at the American Epilepsy Society (AES) annual meeting.
"This study, the largest of its kind to date, provides powerful, replicated real-world evidence to guide individualized, shared decision-making for patients with drug-resistant epilepsy," she said.
Based on the findings, Rai characterized RNS/DBS as "high-risk, high-reward" alternatives to VNS, which has been around much longer and has traditionally been the most popular type of neurostimulation therapy for seizure disorders.
VNS involves placing a pulse generator, generally in the chest or back, with leads running to the vagus nerve in the neck. RNS and DBS differ from each other considerably, but in both cases, the implanted device has leads extending into the brain, and thus require much more invasive surgical procedures.
RNS and DBS now have a substantial evidence base as well, but as Rai noted, no rigorous head-to-head trials have compared them, and none are anticipated. Thus, her group pulled data from the massive Epic Cosmos and TriNetX records storehouses to see how they have stacked up in routine care.
Epic Cosmos holds records for some 300 million individuals in the U.S., while TriNetX covers about half that many in the U.S. and 22 other countries. Among them were about 4.2 million in Epic Cosmos and 1.9 million in TriNetX with diagnoses of epilepsy; 21,375 in the former and 13,860 in the latter had received neurostimulation implants. Patients whose records indicated a Parkinson's disease diagnosis, another condition treated with neurostimulation, were excluded. The researchers also went further, using propensity matching to create cohorts within each database of clinically similar patients treated with the two categories of procedures. In Epic Cosmos there were 5,618 in each group, and in TriNetX there were 2,917.
Serious short-term complications were substantially more common with RNS and DBS. This was most notable for hemorrhagic stroke, which occurred within 3 months in 4% of patients in both databases treated with the brain implants, versus 0.2-0.4% of VNS recipients. Ischemic strokes within 3 months in the Epic Cosmos cohorts were seen in 2.7% with RNS/DBS, compared with 2.0% for VNS; in TriNetX, the corresponding rates were 1.4% and 0.6%, respectively. All of these differences were "highly statistically significant," Rai said.
Another proxy for seizure control was emergency department visits, and this analysis, which covered visits over the 5 years after surgery, favored the more invasive procedures. RNS/DBS were associated with rates of 32% in TriNetX and 49% in Epic Cosmos, versus 41% and 58%, respectively, with VNS.
Overall mortality, meanwhile, was almost identical between procedures through year 5, at which point about 94% of patients were still living, but the tracks then began to diverge. By year 8, the differences were statistically significant: survival was about 91% in the RNS/DBS group compared with 88% for patients receiving VNS. Among patients with 13 years of follow-up, the gap was far wider: 86% versus 76%.
Study limitations included the reliance on administrative data. Also, the propensity matching was made for sex, age, ethnicity, race, comorbidities, smoking, BMI, anti-seizure medications, MRI abnormalities, and epilepsy etiology, but that still left considerable room for unmeasured confounders. And an AES attendee pointed out that RNS and DBS are very different procedures and by lumping them together, it was unclear whether the findings were driven primarily by one or the other.
Disclosures
Rai disclosed no relationships with industry.
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.